Rapid Profiling of the Plasma Proteome and Machine Learning Analytics for Non-Invasive Diagnosis of Alzheimer's Disease

点击翻译按钮获取中文摘要

ABSTRACT The main objective of this project is to develop an innovative blood-based test for highly sensitive and specific, non-invasive and cost-efficient diagnosis of Alzheimer's disease (AD), which would leverage Seer's proprietary Proteograph platform enabled by the convergence of nanotechnology, protein corona, proteomics, and data science. Beyond neuropsychological testing, two approaches have thus far been clinically validated for AD detection, including neuroimaging and analysis of cerebrospinal fluid (CSF)-based biomarkers (e.g., amyloid-β or Aβ). In contrast to the neuroimaging (which is expensive and time-consuming) and CSF analysis (which is less expensive, but involves an invasive lumbar puncture procedure), a blood-based test for AD diagnosis has the potential to be dramatically less costly and easier to implement. Nevertheless, the search for reliable blood- based biomarkers has been challenging and the blood-based detection using ELISA or other epitope-based methods that go after a few biomarkers (e.g., Aβ42 or Tau) have not been successful, presumably owing to the vast dynamic range and high complexity of the plasma components. We have recently demonstrated that our multi-nanoparticle (NP) protein corona technology can facilitate broad and deep profiling of plasma proteome, and by combining with machine learning approaches, could lead to the development of Proteograph classifiers for highly accurate detection of different diseases including AD. As compared to current mass spectrometry- based proteomic techniques that require complex and time-consuming depletion or fractionation workflows for detection of low abundance/rare proteins, our multi-NP protein corona strategy is fast and high-throughput for analysis of the vast body of information in the proteome. In this Direct Phase II project, we will build upon the proof-of-concept studies to further test how Seer's Proteograph platform can be applied to develop a robust blood-based test to detect AD. Specifically, we will identify a panel (~6-10) of NPs from Seer's NP library for broad and deep coverage of the plasma proteome of AD patients (Aim 1); develop Proteograph classifiers and identify the proteins critical for classification through machine learning of the proteomic data generated from the panel of NPs with a cohort of 150 plasma samples of AD and healthy controls (Aim 2); and validate the accuracy of the detection test (based on the important proteins identified in Aim 2) in a separate blind cohort of 450 Aβ-positive AD patients and healthy controls (Aim 3). We expect that the successful completion of this SBIR project will lead to the clinical use of a blood-based AD test, which could further benefit earlier treatment, therapeutic outcomes, and health costs and quality of life for the elderly.